Table 2.

Results for 10-fold cross-validation of logistic discriminant function using quantiles corresponding to the middle 90% of the RMI distribution: odds ratios with 95% confidence limits (CL), P-values based on Fisher’s exact test and adjusted for multiple comparisons.

Percentile (%)	RMI quantile	Odds ratio	Lower CL	Upper CL	False positive error	False negative error	P-valuea
5	−8.5	2.099	0.687	6.455	0.44	0.02	0.52
8	−7.46	2.593	1.025	6.604	0.42	0.03	0.15
12	−6.29	1.618	0.776	3.385	0.40	0.05	0.52
15	−5.53	1.941	0.986	3.838	0.38	0.06	0.19
19	−4.29	2.530	1.351	4.773	0.34	0.07	0.03
22	−3.81	2.139	1.192	3.861	0.33	0.09	0.06
26	−3.06	2.097	1.210	3.655	0.31	0.10	0.05
29	−2.46	2.359	1.381	4.053	0.29	0.11	0.03
33	−1.79	2.192	1.312	3.684	0.27	0.13	0.03
36	−1.09	2.020	1.224	3.352	0.26	0.15	0.05
40	−0.24	1.769	1.084	2.898	0.25	0.18	0.10
43	0.14	1.524	0.940	2.477	0.24	0.20	0.30
47	0.84	1.368	0.847	2.214	0.23	0.23	0.50
50	1.39	1.431	0.887	2.315	0.21	0.24	0.36
53	1.86	1.215	0.752	1.964	0.21	0.27	0.57
57	2.64	1.099	0.678	1.782	0.19	0.30	0.77
60	3.2	1.086	0.666	1.771	0.18	0.32	0.83
64	4.11	1.185	0.719	1.953	0.16	0.33	0.62
67	5.02	1.297	0.779	2.165	0.14	0.34	0.56
71	5.79	1.307	0.769	2.224	0.12	0.36	0.56
74	6.94	1.357	0.783	2.355	0.11	0.38	0.56
78	7.87	1.272	0.711	2.279	0.09	0.40	0.57
81	8.57	1.339	0.723	2.485	0.08	0.42	0.57
85	10.7	1.397	0.706	2.771	0.06	0.44	0.56
88	11.79	1.162	0.553	2.444	0.05	0.46	0.77
92	15.27	0.975	0.400	2.379	0.04	0.49	1.00
95	18.63	0.540	0.172	1.692	0.03	0.51	0.56

^aFisher’s exact test adjusted for multiple comparisons.